These collaborations, which connect expertise across industries and sectors, are aimed at making possible novel solutions to agricultural challenges more quickly and efficiently.
Two collaborations – with IBM Research and with US biotech Maxygen – brought their respective pioneering approaches in data-based predictions modelling, and in the directed evolution of proteins more commonly leveraged in the pharmaceutical industry, together with Syngenta’s world-leading agricultural research and proprietary data sets.
″Helping growers sustainably feed a rapidly growing human population requires a strong collaboration focus, not just across agriculture but across industries,″ said Gusui Wu, Global Head of Seeds Research. ″Collaboration is at the heart of how our scientists approach innovation every day. It is embedded in our scientific culture, and we are continually seeking out different technologies, solutions, and partners to help us better serve farmers.″
IBM Research and Syngenta accelerate optimization of chemical compound synthesis with language models.
Syngenta Group, in collaboration with IBM Research, has enhanced productivity in chemical synthesis using IBM-RXN – a software developed to enable the use of language models for the synthesis of new molecules and materials. IBM-RXN encodes, models, and predicts chemical reactivity. By combining Syngenta’s world-leading chemistry research and proprietary data sets with IBM’s world-class reactivity modelling capabilities and leveraging Natural Language Processing (NLP), IBM’s pioneering modelling approach, enables the partners to deliver scalable, accurate, and data-based predictions modelling. This enables Syngenta to investigate multiple related compounds simultaneously and prioritize routes that offer compounds with the most desirable commercial attributes.
Syngenta is closing the data loop by connecting the IBM-RXN platform to the synthesis platform. Reactivity predictions are now an integral part of the design of synthetic procedures. The models’ outcomes are fully integrated with synthesis planning and execution, establishing a virtual loop where high-quality data generate more relevant models that, in turn, inspire better synthetic procedures. The digitalization of synthetic workflows and the adoption of predictive reactivity modelling are increasing the efficiency and the effectiveness of the synthetic process. Both teams worked on extending reactivity modelling to include bio-catalyzed reactions and metabolic transformations, to support the design of more sustainable synthetic procedures that have a better safety and environmental footprint.
As the predictive power of reactivity models increases, scientists may become increasingly confident in delegating part of their work to AI-enabled automation. This should allow shifting the focus to the synthesis strategy and overall chemical design.
